Do you have to sequence both strands?

Lisa Vawter vawter at
Tue Apr 12 15:10:43 EST 1994

In article <01HARUDG2ENK005IEC at>,
>Dear Netters;
>    While I agree in general that compressions require sequence to be determined
>on both strands to have any confidence that you've got it right, I would
>like to present a situation where I do not think it is necessary to
>always sequence both strands.  We are a molecular systematics lab, and we have
>just finsished sequencing the mitochondrial COII gene from 19 species of
>black flies.  In all the species, the gene is 690 base pairs in length, 
>it doesn't have any particularly trecherous regions of compression, and its
>amino acid sequence is quite conserved across taxa.  If I were to obtain nice
>looking, completely overlapping sequence from only one strand of the next 
>black fly species, and the sequence maintained reading frame, aligned 
>sensibly (conservative replacements) and was 690 base pairs in length, I would
>deem it a waste of time and (ultimately) taxpayers' money to sequence the other
>strand. Assuming replicated reading and no ambiguities, the chances that a
>compression has fooled me, while retaining the above characteristics seem
>to me to be very low, and the chances that such an error would cause me to
>draw incorrect phylogenetic conclusions are just as low again.  Sure, it is 
>possible, but it is also possible that a combination of unlikely artifacts
>introduces errors into sequence determined from both strands.
>     I should note that almost all of our sequences have been determined 
>completely from both strands, .....

(snip, snip)

I, too, sequence to gather data for phylogenetic analysis.  I 
respectfully disagree with you for the following two reasons:  (1)You 
state that nice-looking sequence with no ambiguities need not be 
sequenced from both directions.  Unfortunately, ambiguities are often not 
obvious, even in nice-looking sequence, except for causing disagreement 
when the second strand is sequenced.  (2) More important, many sequencing
errors are non-random.  They are due to minor variations in the 
sequencing reactions and to idiosyncrasies of various bits of 
template.  Read **APPARENT SYNAPOMORPHY**.  This is, to me, the 
scariest part of using nucleic acid sequence data for phylogenetic
analysis.  We must be especially vigilant of the quality of our
sequence data, not less careful, because we are using it for this

Lisa Vawter
Harvard University
vawter at

More information about the Methods mailing list